AI in Schools: The Binding Safety and Privacy Standard Microsoft and Anthropic Signed With Teachers Unions (and 10 Questions to Ask Your School)
By Equipo IA University Kids · Published · Updated
Leer este artículo en españolMinimum age: 18
What's new
On November 10, 2026, the American Federation of Teachers (AFT) and the United Federation of Teachers (UFT) announced a legally binding agreement with Anthropic, the company behind Claude. It makes the AI Safety and Privacy Standard for Schools enforceable for school districts using Claude for Teachers, and it automatically applies to Anthropic's existing agreements with those districts.
Anthropic isn't the first. According to GovTech, Microsoft adopted the standard in September, and its official fact sheet says that starting November 1, 2026, any U.S. school district can add these protections to new or existing Microsoft agreements without renegotiating.
The 10 protections in plain English
Based on Microsoft's fact sheet and the signed agreement:
- No training AI on student or educator data (with a narrow safety and security exception).
- Data minimization: tracking, profiling, biometrics and AI "memory" face limits and review.
- Schools and students own their data. It can't be sold or used for ads, and schools control retention and deletion.
- AI supports, people decide. High-risk decisions need human review, and AI companions are prohibited.
- Real accountability: certifications, breach notification within 72 hours, and fixes, or AI use is paused or ended.
- Security that matches the stakes: encryption, access controls and independent testing.
- Transparency for families: plain-language guides explaining how the products work and what data they collect.
- Equity and accessibility for every student.
- No feature creep or vendor lock-in: schools get notice of major new features and can take their data elsewhere.
- Long-term responsibility: bans on training with or selling the data continue after the contract ends.
The agreement is especially direct about relationship-style chatbots. Providers may not offer companion-style or relationship-oriented features designed to foster emotional attachment or dependency, keep students engaged longer than the learning task requires, simulate friendship or encourage students to share sensitive information. It also requires crisis protocols for content about self-harm, suicide or violence.
What it doesn't change
- Claude for Teachers is for educators only. The Claude Help Center describes it as free for verified U.S. K-12 educators through July 30, 2028, and states plainly that it is not for students. Anthropic's Consumer Terms also require users to be at least 18.
- As GovTech notes, that means the provisions on students' direct use of AI don't currently apply to Anthropic, although student information that teachers enter is still protected.
- It's not a law. It's a voluntary, contract-based standard for U.S. districts, so it doesn't apply automatically outside the United States.
Why it matters for parents and teachers
Schools are adopting AI tools quickly, and families often don't know what data about their kids is shared or why. This standard gives you something practical: a concrete list of what families can reasonably expect. In the AFT announcement, AFT President Randi Weingarten said the goal is that no teacher or parent should have to choose between AI tools and children's safety and privacy.
The same release notes that in May 2026 the AFT also called for a ban on student-facing AI in elementary school and on "social companion" chatbots before age 16.
For parents and teachers: 10 questions to ask your school
Use the standard as a guide to talk with your school about any AI tool used with your kids or students:
- Which AI tools do teachers use, and which do students use directly?
- Does the vendor use student data to train its AI?
- What data is collected, who owns it, and when is it deleted?
- Is any of it sold or used for advertising?
- Does AI make any decision about my child (grades, supports, discipline) without a person reviewing it?
- Does the tool have companion or "friend" features?
- What happens if a student types something about self-harm or violence?
- How quickly will families be told about a data breach?
- Is there a plain-language guide for families?
- Who can we contact with questions?
For teachers: go through these questions with your administrators before bringing a tool into class. With students (around age 10 and up), try reading the privacy policy of an app they use and marking which questions it answers and which it doesn't.
Sources
- AFT, UFT and Anthropic Sign Binding AI Safety and Privacy Agreement for Schools (American Federation of Teachers, press release) · 2026-11-10
- AI Safety & Privacy Standard for Schools - Memorandum of Agreement, Version 1.0 (signed agreement, National Academy for AI Instruction) · 2026-10-22
- Microsoft Fact Sheet: AI Safety & Privacy Standard for Schools · consultado 2026-11-15
- Anthropic Joins AFT and UFT's Binding AI Safety Standard (GovTech) · 2026-11-11
- Claude for Teachers: your data and our terms (Claude Help Center) · consultado 2026-11-15
- Anthropic Consumer Terms of Service (minimum age) · consultado 2026-11-15